Patch for taken from . --- trunk/src/pcre2_internal.h 2016/11/19 12:46:24 600 +++ trunk/src/pcre2_internal.h 2017/02/24 18:25:32 670 @@ -1774,10 +1774,17 @@ /* UCD access macros */ #define UCD_BLOCK_SIZE 128 -#define GET_UCD(ch) (PRIV(ucd_records) + \ +#define REAL_GET_UCD(ch) (PRIV(ucd_records) + \ PRIV(ucd_stage2)[PRIV(ucd_stage1)[(int)(ch) / UCD_BLOCK_SIZE] * \ UCD_BLOCK_SIZE + (int)(ch) % UCD_BLOCK_SIZE]) +#if PCRE2_CODE_UNIT_WIDTH == 32 +#define GET_UCD(ch) ((ch > MAX_UTF_CODE_POINT)? \ + PRIV(dummy_ucd_record) : REAL_GET_UCD(ch)) +#else +#define GET_UCD(ch) REAL_GET_UCD(ch) +#endif + #define UCD_CHARTYPE(ch) GET_UCD(ch)->chartype #define UCD_SCRIPT(ch) GET_UCD(ch)->script #define UCD_CATEGORY(ch) PRIV(ucp_gentype)[UCD_CHARTYPE(ch)] @@ -1834,6 +1841,9 @@ #define _pcre2_default_compile_context PCRE2_SUFFIX(_pcre2_default_compile_context_) #define _pcre2_default_match_context PCRE2_SUFFIX(_pcre2_default_match_context_) #define _pcre2_default_tables PCRE2_SUFFIX(_pcre2_default_tables_) +#if PCRE2_CODE_UNIT_WIDTH == 32 +#define _pcre2_dummy_ucd_record PCRE2_SUFFIX(_pcre2_dummy_ucd_record_) +#endif #define _pcre2_hspace_list PCRE2_SUFFIX(_pcre2_hspace_list_) #define _pcre2_vspace_list PCRE2_SUFFIX(_pcre2_vspace_list_) #define _pcre2_ucd_caseless_sets PCRE2_SUFFIX(_pcre2_ucd_caseless_sets_) @@ -1858,6 +1868,9 @@ extern const uint32_t PRIV(vspace_list)[]; extern const uint32_t PRIV(ucd_caseless_sets)[]; extern const ucd_record PRIV(ucd_records)[]; +#if PCRE2_CODE_UNIT_WIDTH == 32 +extern const ucd_record PRIV(dummy_ucd_record)[]; +#endif extern const uint8_t PRIV(ucd_stage1)[]; extern const uint16_t PRIV(ucd_stage2)[]; extern const uint32_t PRIV(ucp_gbtable)[]; --- trunk/src/pcre2_ucd.c 2015/07/17 15:44:51 316 +++ trunk/src/pcre2_ucd.c 2017/02/24 18:25:32 670 @@ -41,6 +41,20 @@ const char *PRIV(unicode_version) = "8.0.0"; +/* If the 32-bit library is run in non-32-bit mode, character values +greater than 0x10ffff may be encountered. For these we set up a +special record. */ + +#if PCRE2_CODE_UNIT_WIDTH == 32 +const ucd_record PRIV(dummy_ucd_record)[] = {{ + ucp_Common, /* script */ + ucp_Cn, /* type unassigned */ + ucp_gbOther, /* grapheme break property */ + 0, /* case set */ + 0, /* other case */ + }}; +#endif + /* When recompiling tables with a new Unicode version, please check the types in this structure definition from pcre2_internal.h (the actual field names will be different):